How this instrument works
HOMA stands for homeostasis model assessment — a mathematical description of the feedback loop between the liver, the pancreas, and circulating blood sugar, published in 1985 by Matthews, Hosker, Rudenski, Naylor, Treacher, and Turner. Multiply fasting glucose in mg/dL by fasting insulin in µU/mL, then divide by 405; that constant scales the product down to match the resistance values the team measured directly with hyperinsulinemic clamp studies in the same volunteers.
A higher result means the pancreas is releasing more insulin than a textbook amount to hold that fasting glucose level where it sits — muscle, fat, and liver tissue are resisting the hormone's signal, and beta cells are compensating by working harder. A lower result means the same glucose level is being maintained with comparatively little insulin, evidence the tissues are responding readily. Because the whole calculation needs only a single blood draw rather than hours of continuous glucose infusion, it has become the everyday stand-in for the clamp method it was built to approximate.
There is no single diagnostic line the field agrees on. Cutoffs cited for calling someone resistant range roughly from 1.9 to 2.9, shifting with the population studied and the specific assay a lab uses to measure insulin. Treat the number as one data point read alongside fasting glucose, A1C, and the rest of a metabolic panel — not a verdict delivered on its own.
- Draw the two values from a true fasting blood panel — 8 to 12 hours with nothing but water beforehand.
- Enter Fasting glucose in mg/dL.
- Enter Fasting insulin in µU/mL.
- Read HOMA-IR; the working block shows the multiplication and the division by 405.
Worked example — three fasting panels
Fasting glucose 90 mg/dL, fasting insulin 10 µU/mL: 90 × 10 = 900, divided by 405 gives a HOMA-IR of 2.22 — right at the edge of the ranges some labs flag as resistant, though not a hard boundary on its own.
Push glucose to 126 mg/dL (the threshold used for a diabetes diagnosis by itself) with insulin at 25 µU/mL: 126 × 25 = 3150, divided by 405 gives 7.78, a figure that lands well inside resistant territory under any commonly cited cutoff.
A leaner pairing — glucose 85 mg/dL, insulin 5 µU/mL — gives 85 × 5 = 425, divided by 405 equals 1.05, comfortably inside ranges considered sensitive.
Questions
What counts as a normal HOMA-IR?
There isn't one agreed number. Values under roughly 1.0-1.5 are usually read as sensitive, and thresholds for calling someone resistant are cited anywhere from about 1.9 to 2.9 depending on the population studied and the specific insulin assay a lab runs. Read the figure alongside the rest of a metabolic panel rather than as a stand-alone verdict.
Does the constant change if glucose is reported in mmol/L?
Yes. The formula as published uses mg/dL for glucose and µU/mL for insulin, giving the 405 divisor. Labs that report glucose in mmol/L (common outside the US) use an equivalent version dividing by 22.5 instead. Converting glucose to mg/dL first (multiply mmol/L by 18) and sticking with 405 works just as well — just don't mix the wrong constant with the wrong unit.
Why does this only need one blood draw when clamp studies take hours?
The clamp method infuses glucose and insulin continuously over two or more hours to directly measure how much sugar the body clears at a fixed hormone level — precise, but impractical outside a research setting. HOMA reversed the approach: it modeled the fasting feedback loop mathematically and solved for a resistance figure using two values a routine fasting panel already produces, then validated that shortcut against clamp results in the original cohort.
Can this number diagnose diabetes or PCOS by itself?
No. It is a screening and research tool, not a diagnostic criterion. Diabetes is confirmed through fasting glucose, A1C, or an oral glucose tolerance test; PCOS relies on the Rotterdam criteria. Insulin resistance shows up often in both, which is why this figure gets measured alongside those workups, but one elevated reading without supporting findings isn't a diagnosis.
Does the length of the fast change the result?
Considerably. The formula assumes a genuine fasting state, typically eight to twelve hours with only water allowed. Eating, sweetened coffee, or even the mild insulin effect of caffeine shortly before the draw can shift glucose and insulin enough to move the ratio meaningfully — one reason labs enforce a strict fasting window before drawing either value.
Is there a companion figure for how well the pancreas is secreting insulin?
Yes — the same 1985 model also produces HOMA-B, an estimate of beta-cell function built from the identical fasting glucose and insulin pair, using a different formula. HOMA-IR and HOMA-B are companion outputs of one framework: one describes resistance out in the tissues, the other describes how hard the pancreas is pushing to overcome it.
References
- Matthews DR et al. 1985, Diabetologia — original HOMA model (PubMed)
- NIDDK — Insulin Resistance & Prediabetes
Read this first: This instrument computes a screening figure from population formulas — it is not a diagnosis, and it cannot see the whole picture a clinician can. Use it to inform a conversation, not to replace one.